Llms

Activation Oracles: Training and Evaluating LLMs as General-Purpose Activation Explainers

Activation Oracles are LLMs trained to read activation snapshots and answer natural-language questions, generalizing to misalignment audits and secret-elicitation tasks.

Read More →

Concept Poisoning: Probing LLMs without probes

A novel LLM evaluation technique using concept poisoning to probe models without explicit probes

Read More →